The Aquacrop-OSPY model was used to generate a large number of runs varying irrigation schedules and depths. The first conlusion of this paper is that as documented in the literature in the introduction, Aquacrop-OSPY is a powerful tool that extends the useability of the standard Aquacrop package, although it does have some downsides. Due to the recent release of the package, the relatively small number of researchers implementing the package and the reliance on an individual rather than an organization to maintain it leads to some shortcomings, such as the inability to implement sprinkler irrigation rather than targeted irrigation in this report, and a much smaller documentation compared to the original package.
These shortcomings, however are more than made up for by the benefits of the package. As an open source package, it can be run on any operating system. Further, moving to matlab/octave allows users to see and amend the underlying relationships between parameters to keep the Aquacrop package relevant as research into crop modeling expands beyong what the original creators could have seen in the release of Aquacrop 15 years ago. Most importantly, it allows the user to easily generate a large number of runs and compile the data programmatically and create irrigation management functions to optimize irrigation dynamically.
The second conclusion of this report is that varying the dates and depths of irrigation has significant effects on all crop growth outputs in the aquacrop model. There is a direct relationship between increasing irrigation depth and increasing all crop growth parameters and there is an inverse relationship between irrigation depth and the variance of the crop growth parameters. In adiditon, nearly all of the chosen intervals outperformed the random baseline at a 10 mm depth.
In order to extend this research there are several fruitful avenues. The first would be to amend the irrigation management method within the Aquacrop-OSPY package to easily allow sprinkler and surface flood irrigation, just as most other parameters are easily accessed and modified. A second fruitful avenue would be to perform a properly interpreted statisitical analys in consultation with field experts, for example a multiple regression analysis and fitting a 3d function generated by the data